ORIGINAL ARTICLE

Interpretation biases in chronic pain patients: an incidental learning task A. Khatibi1,2, L. Sharpe3, H. Jafari2,4, S. Gholami2, M. Dehghani5 1 2 3 4 5

Laboratory of Research on Neuropsychology of Pain, University of Montreal, Canada Rehabilitation Research Centre, Rehabilitation Sciences Faculty, Iran University of Medical Sciences, Tehran, Iran School of Psychology, Clinical Psychology Unit F12, The University of Sydney, Australia Research Group on Health Psychology, University of Leuven (KU Leuven), Belgium Family Research Institute, Shahid Beheshti University (G.C.), Tehran, Iran

Correspondence Ali Khatibi E-mail: [email protected] Funding sources This study (data collection to manuscript preparation) was supported by the Rehabilitation Research Centre of Iran University of Medical Sciences (Grant No. 17123). The contribution of Ali Khatibi was supported by a grant from the Canadian Institute of Health Research (CIHR MOP130341). Conflicts of interest None declared. Accepted for publication 30 October 2011 doi:10.1002/ejp.637

Abstract Background: The aim of this study was to investigate the impact of chronic pain on interpretation bias for ambiguous faces, using a recently developed paradigm with ecologically valid stimuli. Methods: Fifty patients with chronic pain and 25 healthy controls were trained to respond to probes following the presentation of happy or painful faces, using an incidental learning task. During a test phase, ambiguous faces were presented. The degree to which participants were faster to respond to probes presented where painful (rather than happy) faces had previously been presented was taken as an indication of the interpretation bias towards painful faces. Results: All participants had learnt the originally presented contingency. As predicted, chronic pain patients showed a greater bias towards interpreting ambiguous faces as painful than control participants. Further, there were correlations between fear of pain and catastrophizing and interpretation bias, indicating that participants with higher fear of pain and higher scores on a measure of catastrophizing were more likely to interpret ambiguous faces as painful. Severity of pain was inversely associated with increased interpretation bias for pain. Conclusion: These results show clear evidence that chronic pain patients do demonstrate an interpretation bias towards painful faces and that this bias is greater for those who catastrophize more and have higher levels of fear of pain, but experienced less pain in the preceding week. Given the recent potential shown for interventions that modify cognitive biases, this paradigm would seem to be well suited to future efforts to modify interpretation biases in pain.

1. Introduction Theories of chronic pain emphasize the importance of attentional processes, like hypervigilance (e.g., Vlaeyen and Linton, 2000) and difficulty disengaging (e.g., Eccleston and Crombez, 1999). However, recent re-conceptualizations (Eccleston and Crombez, 2007; Crombez et al., 2012) have more thoroughly articulated the processes underlying the capacity of pain to © 2014 European Pain Federation - EFICâ

capture attention. The misdirected problem-solving model suggests that when individuals interpret pain as a problem needing a solution, the individual will be motivated to avoid pain (Eccleston and Crombez, 2007). Hence, the interpretation of pain as harmful is thought to increase the propensity that individuals will interpret ambiguous stimuli as pain related. Further, as the interpretation of pain as harmful increases fear of pain and the tendency to catastrophEur J Pain 19 (2015) 1139--1147

1139

Interpretation bias in chronic pain

What’s already known about this topic? • Chronic pain patients show an interpretation bias to word-related pain stimuli on a variety of paradigms. What does this study add? • Chronic pain patients interpreted ambiguous faces as pain-related in comparison to controls. • Interpretation bias towards pain was strongly associated with pain catastrophizing and inversely related to pain experienced in the previous week. • This paradigm, using ecologically valid stimuli, has the potential to be used to modify interpretative biases.

A. Khatibi et al.

et al., 2014), most studies have modified attention, despite the less robust evidence for attention than interpretation biases. CBM for interpretation (CBM-I) has so far only been applied in the laboratory (Jones and Sharpe, 2014), although has shown promise. Hence, the validity of ecologically valid methods to assess interpretive biases could potentially have clinical applications. We predict that using our incidental learning paradigm, chronic pain patients will demonstrate interpretation biases towards pain compared with controls. In addition, we hypothesize that the interpretation biases will be associated with theoretically relevant variables, such as fear of pain and catastrophizing.

2. Methods ize about pain. With pain reduction as the primary goal, the cycle of being fearful, catastrophizing, overattending to pain and avoiding activities continues, leading to increased disability (Eccleston and Crombez, 2007; Crombez et al., 2012). Hence, the interpretation of pain as harmful is crucial to the cycle that develops and perpetuates pain. That is, it is interpretations that are thought to lead to individuals to avoid normal activities that would promote recovery. Surprisingly, interpretation biases in pain have been considerably less researched than other cognitive biases, such as attentional biases (e.g., Schoth et al., 2012; Crombez et al., 2013). Only five studies have investigated interpretation biases in pain patients (Edwards and Pearce, 1994; Pincus et al., 1994, 1996; Griffith et al., 1996; McKellar et al., 2003). All studies found evidence that chronic pain patients were more likely to interpret ambiguous stimuli as pain related using different paradigms, including ambiguous homonyms (Pincus et al., 1994), homographs (McKellar et al., 2003), homophones (Pincus et al., 1996) and word-stem completion (Edwards and Pearce, 1994; Griffith et al., 1996). Hence, interpretation biases in pain patients appear robust, but all paradigms used to date have relied on word stimuli, known to be less ecologically valid than pain-related faces. In the anxiety disorders, where interpretation biases are also found, methods of modifying these biases have been developed, known as cognitive bias modification (CBM). Reviews and meta-analyses have confirmed the therapeutic potential of CBM in anxiety (e.g., Hakamata et al., 2010; Hallion and Ruscio, 2011; Beard et al., 2012; MacLeod and Mathews, 2012). In the pain literature, although CBM has shown promise (McGowan et al., 2009; Sharpe et al., 2012; Schoth 1140 Eur J Pain 19 (2015) 1139--1147

2.1 Participants The study was conducted at three physiotherapy clinics in Tehran, Iran. Inclusion criteria were (1) persistent pain in lower part of back for more than 3 months; (2) being over 18 years old; and (3) having sufficient literacy to complete questionnaires. Exclusion criteria were (1) history of head injury in the past 2 years; (2) using psychotropic medication; and or (3) inability to work with response box and computer. The study was conducted in accordance with Helsinki Declaration ethics in research and the study was approved by the ethics committee of the Faculty of Rehabilitation Science, Iran University of Medical Sciences. The sample size was calculated based on an effect size (Cohen’s d) of 0.66 in previous studies (e.g., Pincus et al., 1996). According to G-power (Version 3.1. http://www.gpower.hhu.de/en.html) software, we required a total of 73 participants to achieve 80% power with an alpha level of 0.05. Eighty-four consecutive chronic low back pain patients were invited to take part in this study and 50 patients (59.5%) agreed (mean age = 43.6 years, SD = 9.8; range = 24–69; 38% male). Pain duration was between 7 and 60 months (mean = 25.7 months; SD = 16.0). All patients were on the waiting list for a 10-session routine physiotherapy treatment. A non-clinical group of 25 painfree individuals (44% male), of similar age and educational level, also participated in the study (they were recruited through flyers in nearby social places; mean age = 36.2 years, SD = 6.2, range = 26–51). The exclusion criteria for the patient group also applied to the control group, and in addition, a history of chronic pain was an exclusion criterion for this group.

2.2 Measurements Participants were asked to complete a battery of questionnaires in order to assess their level of pain, anxiety, stress, depression, pain-related fear, pain catastrophizing and dis© 2014 European Pain Federation - EFICâ

A. Khatibi et al.

ability reported due to pain. The Persian version of all measures described below are used in previous studies and reported to have good psychometric properties in a similar sample populations (Khatibi et al., 2009; Mohammadi et al., 2012).

2.2.1 Pain severity: Visual Analogue Scale (VAS) The VAS is a 10-centimetre graded line with two anchors. The left anchor indicates ‘no pain at all’ and the right anchor indicates ‘extremely intense pain’. Participants were asked to the rate the average intensity of their pain in the week prior to testing on the VAS.

2.2.2 Pain Catastrophizing Scale (PCS; Sullivan et al., 1995) The PCS consists of 13 items describing the degree to which they tend to predict negative outcomes (i.e., catastrophes) when in pain. Participants indicate the degree to which they have each particular catastrophic thought on a 5-point Likert scale (0 = not at all; 4 = all the time). Higher total scores reflect higher levels of pain catastrophizing. The measure has shown good psychometric properties in both clinical and

Interpretation bias in chronic pain

non-clinical populations (Sullivan et al., 1995). The Cronbach’s alpha in this sample was 0.86.

2.2.3 Fear of Pain Questionnaire (FPQ; McNeil and Rainwater, 1998) The FPQ is a 30-item questionnaire designed to an evaluate individual’s level of fear related to pain on a 5-point Likert scale ranging from (1 = not at all; 5 = extreme). This questionnaire has shown good psychometric properties in both clinical and non-clinical populations (Roelofs et al., 2005). For the current sample of participants, the Cronbach’s alpha was 0.87.

2.2.4 Incidental learning task The incidental learning task (cf. Yoon and Zinbarg, 2008) consisted of two main phases: a learning phase and a testing phase (Fig. 1). During the learning phase, painful and happy expressions were presented individually on the centre of the screen along with two target locations on the left- and righthand side of the expression, followed by a target (e.g., H) which appeared immediately following the faces at one of two predefined locations (i.e., left or right). Participants were

Figure 1 Pictorial representation of typical trials in the learning phase (left) and testing phase (right) of the incidental learning task. This specific example is taken from the task version focusing on painful versus happy expressions.

© 2014 European Pain Federation - EFICâ

Eur J Pain 19 (2015) 1139--1147

1141

Interpretation bias in chronic pain

required to indicate the location of the target. Expression type (happy or painful) predicted the target’s location on 80% of the learning trials. Hence, participants were expected to learn the association between a painful or happy face and the target location. During the test phase, morphed facial expressions were presented and followed by the target, randomly presented in each of the two previously presented positions (i.e., left and right). Because participants have learnt an association (e.g., happy faces are followed by a target on the right and painful faces are followed by a target on the left), the degree to which participants respond more quickly to targets presented where they had previously followed a painful face compared with targets in the other location is indicative of the degree of interpretation bias.

2.2.4.1 Learning phase (Fig. 1; left panel) The incidental learning task was originally developed to indirectly evaluate individuals’ interpretation of ambiguous facial expressions (Yoon and Zinbarg, 2008). Yoon and Zinbarg (2008) used negative and positive facial expressions and we substituted pain-related and happy facial expressions. The only difference between the paradigms other than the stimuli was that we simplified the task, such that instead of four target locations used in their study, we used a modified version of task with two target locations on the right and left side of the fixation cross (Khatibi et al., 2014). During the learning phase, each trial started with a black central fixation cross on a grey background and two square position markers (black frames, 1 × 1cm), one at the left and one at the right of the fixation cross. The inner edge of the target position marker distanced 12 cm (horizontal axis) from the fixation cross (500 ms). The cross was replaced by a happy or painful facial expression (675 ms), immediately followed by a target (‘H’, 0.85 × 0.85 cm). For half of the participants, (1) happy expressions were followed by a target at the left side of the fixation cross in 80% of the trials (i.e., location predicted by happy expressions, not by painful expressions) and at the right side in 20% of the trials (i.e., location not predicted by happy expressions, but by painful expressions); and (2) painful expressions were followed by a target at the right of the fixation cross in 80% of the trials (i.e., location predicted by painful expressions, not by happy expressions) and at the left side in 20% of the trials (i.e., location not predicted by painful expressions, but by happy expressions). For the other participants, predicted location was counterbalanced. The participants’ task was to indicate the target’s position as quickly and accurately as possible, by pressing the corresponding key on the response box. As soon as the response was given, or after 3000 ms, the screen was refreshed and the next trial began after an inter-trial interval of 800–1200 ms. The learning phase consisted of two blocks, each consisting of 32 trials (16 happy and 16 painful expressions). Photographs of each facial expression were presented twice, once during each trial block. Trials were presented in a different random order for each participant. 1142 Eur J Pain 19 (2015) 1139--1147

A. Khatibi et al.

2.2.4.2 Test phase (Fig. 1; right panel) The test phase was similar to the learning phase, except that 16 ambiguous expressions were presented, followed by a target presented equally often on the left or the right. Interspersed with the ambiguous faces, were eight of the original happy and painful expressions that were presented during the learning phase (always followed by a target at the expected target location). These eight faces were randomly (pre)selected from the stimuli presented in the learning phase. The reason for continuing to present the happy/ painful faces during the test phase was to ensure that the learning was consolidated and retained throughout the test phase. Each photograph of an expression was presented once. Trials were presented in a different random order for each participant. Previous research has questioned the internal consistency of some of the experimental paradigms used to measure attentional bias (e.g., Stroop or dot-probe; see Dear et al., 2011). Cronbach’s alphas were calculated for the incidental learning task based on the reaction times of participants to the ambiguous test phase trials. These showed excellent internal consistency of this paradigm (Cronbach’s alpha = 0.876). Indeed, when reaction times of all trials were considered, the Cronbach’s alpha improved further (Cronbach’s alpha = 0.948), and even for individual phases and stimuli, the Cronbach’s alphas were acceptable (Cronbach’s alphas = 0.629–0.848).

2.2.5 Face stimuli From a previously validated database (Khatibi et al., 2014), 16 happy, 16 painful and 16 ambiguous painful/happy (morphed between happy and painful: showing 50% pain– 50% happiness in the face) were used for the purpose of this study. All pictures were presented to 10 experts familiar with facial action coding system and were selected according to ratings provided by those experts. Colour pictures sized (4.5 × 6 cm) were used. Non-facial features were removed and replaced with a uniform grey background because this information might distract participants from the task of processing the expression (Nusseck et al., 2008). An example of morphed and original expressions is provided in Supporting Information Fig. S1.

2.3 Procedure The patients were approached by the experimenter (S.G.) and the aim of study was explained to them. Those who volunteered to participate in the study gave written consent and were directed to a separate room to complete the task. Participants were then given instructions about the incidental learning task. They were told that a series of faces would be presented and that their task was localize the target that followed these faces as quickly as possible. They were also informed that there would be an association between one aspect of faces and the location of the target and that they © 2014 European Pain Federation - EFICâ

Interpretation bias in chronic pain

A. Khatibi et al.

needed to recognize the pattern and report it to the experimenter at the end of session. They were also shown how to operate the response box and when they had no further questions, the incidental learning task was commenced. Because there were two learning phases, no practice trials were included. After each phase, participants were given the opportunity to ask any questions regarding the task. Once the entire task had been completed, participants were asked to identify the association that they had learned and to complete the battery of questionnaires. At the end of session, participants were debriefed and left the testing room. The procedure was identical for both groups of participants, except that the control participants did not receive the pain intensity rating.

2.4 Analyses Initially, in order to confirm that the participants did learn the association, we conducted a 2 (target type: happy vs. painful) × 2 (target location: as predicted by happy or painful) analysis of variance (ANOVA). Hence, an interaction effect would confirm that participants responded more quickly to targets as predicted by the relevant facial expression and, hence, that the contingency was indeed learned. We also conducted a 2 (target type) × 2 (target location) × 2 (group: chronic pain vs. healthy control) ANOVA to determine whether the learning effect differed between the groups. The main outcome analyses to determine whether chronic pain patients demonstrated interpretive biases were a 2 (group: chronic pain vs. healthy control) × 2 (target location: as predicted by happy vs. painful face) mixed-model ANOVA. In addition, we calculated an interpretation bias index, where the reaction time when the target was in the location predicted by the happy face was subtracted from the reaction time when the target was in the location predicted by the painful face. Hence, a negative value indicated a bias towards painful faces. This bias index was then used in correlations with variables of interest, namely pain level, catastrophizing and fear of pain. Correlations were provided separately for the chronic pain and healthy control group.

3. Results 3.1 Data preparation Due to the low rate of incorrect responses (1.56% total responses), analyses were only performed on reaction times. For each participant, incorrect responses and responses deviating 2.5 SD from the mean reaction times (RTs) of that subject were removed from the analysis (less than 1% of trials). Mean RTs were calculated: for each subject, for each type of expression and for each target location. An interpretation index was calculated, as follows. The mean RT to targets presented at the location predicted © 2014 European Pain Federation - EFICâ

by happy expressions on trials with ambiguous faces were subtracted from the mean RTs to targets at the location predicted by painful expressions following ambiguous expressions. Hence, negative interpretation indices represent negative (painful) interpretation of ambiguous expressions.

3.2 Learning phase In order to confirm that the learning phase was successful and that participants reliably learnt the associations between happy faces and the target location and painful faces and the target location, the following analyses were conducted. Mean RTs during the learning phase were analysed using a repeated measures ANOVA with target location (2: target at the location predicted by painful expressions vs. target at the location predicted by happy expressions) and expression (two facial expressions: happy vs. painful) as withinsubjects factors and group (2: chronic pain vs. control) as the between-subjects factors. There was a significant two-way interaction between the target location and facial expression F(1,73) = 65.13; p < 0.001; 2 = 0.47 . Further, paired sample t-tests showed that following painful expressions, participants reacted to targets at the location predicted by painful expressions (823.82 ± 240.5) significantly faster than they reacted to targets at the location predicted by happy expressions (971.73 ± 239.3), t(74) = 6.08; p < 0.001; Cohen’s d = 0.6; 0.3 < 95% CI for effect size < 0.9). Similarly, following the presentation of happy expressions, participants reacted to targets at the location predicted by painful expressions (937.55 ± 183.7) significantly more slowly than their reactions to targets at the location predicted by happy expressions (812.09 ± 260.5), t(74) = 4.26; p < 0.001; Cohen’s d = 0.5; 0.2 < 95% CI for effect size

Interpretation biases in chronic pain patients: an incidental learning task.

The aim of this study was to investigate the impact of chronic pain on interpretation bias for ambiguous faces, using a recently developed paradigm wi...
2MB Sizes 4 Downloads 4 Views